The present invention relates to multi-mainframe computing systems, and more specifically, to systems and methods for serialization of access to data resources in multi-mainframe or heterogeneous computing environments.
In a typical multi-mainframe computing environment, a mainframe can include multiple servers. For example, in a System z hybrid environment, an IBM® System z mainframe (e.g. zEnterprise® 196) is coupled to IBM® blades housed in one or more IBM® BladeCenter® instances. The coupling is accomplished by Ethernet networks via Transmission Control Protocol/Internet Protocol (TCP/IP). If any data needs to be transferred from the mainframe to a server, the data passes over the Ethernet network. Applications involving high volume and high speed data transfers, such as database accesses (e.g., via Direct Access Storage Devices (DASD)) within the multi-mainframe computing environment can include considerable processing that may impact other work in a mainframe.
With a suitable channel implementation on the server coordinating the data transfers, it is possible for the server and mainframe to share access to the same database storage devices (e.g., DASD). As such, the server can directly access the data and load the data without requiring much mainframe processing or transfer over the Ethernet network.
In multi-mainframe operating systems that manage concurrently hosted units of work, serialization (e.g., Global Resource Serialization (GRS)) is a mechanism that ensures that multiple conflicting accesses to a data set are serialized so that the integrity of the data is preserved. For example, if multiple users were allowed to update the same data set at the same time, data corruption could occur. Serialization allows users that update a data set to obtain exclusive access to the data set while they are updating it. However, sometimes there are servers that are not part of the multi-mainframe operating system, but still do access a common database shared between the multi-mainframe operating system servers and the external computers. In such a case, the serialization service chosen to be used by the resource owner on the multi-mainframe operating system may not be available to the external servers. In this situation there is no way to serialize accesses to data set resources among programs running on the multi-mainframe operating systems and programs running simultaneously on the external server operating systems.